Fundamentally, a pixel is a unit to describe a 2D image and has X and Y coordinates. The number of pixels in a certain field of view can be described as resolution. So, a pixel is a unit that makes up 2D digitised images.

While voxels are three dimensional and instead of X and Y, they have an additional dimension which is Z. Therefore, instead of a 2D square you can think of a voxel as a 3D cube. For image processing this means that instead of working on a plane we then work with a volume described by voxels.

Depending on the data you work with you will want to work with 2D or 3D data comments such as for example X ray is typically 2D and MRI is typically 3D data. In biomedical imaging data are often acquired in 3D but actually analysed in 2D.

2D vs 3D Filters

When working with 3D data you also need to think about 3D filters for example. Typically,2D filters are limited to a single plane, while 3D filters operate in the volumetric space (i.e. consider the neighbouring slices or planes). Thus, when working with 3D data, applying a 2D filter will only work in a single slice at a time.

3D is Often More Challenging in Image Analysis

Working with 3D voxels includes more data (instead of X Y, it is X Y * Z) and steps such as filtering need to consider a whole new dimension, which increases the complexity. Applying algorithms developed for 2D to 3D data typically does not work, and therefore adaptions need to be made.

Also, it is important to consider that voxels aren’t the only way to think about 3D data, for example, point clouds or frequency domains can be used to describe data. Whatever domain you choose, each domain has its pros and cons for understanding spatial information

If you want a free guide to image analysis resources, click here.

If you want to learn more about biomedical image analysis, then click here.